Invited speakers
Mark Sanderson
Getting Rid of the Ten Blue Links
In this talk, I will describe the work we are doing at RMIT to change one of the commonest web pages we all look at: the Search Result Page (SERP). In our work we are looking to replace the SERP with a set of answer passages that address the user’s query. In the context of general web search, the problem of finding answer passages has not been explored extensively. Previous studies have found that many informational queries can be answered by a passage of text extracted from a retrieved document, relieving the user from having to read the actual document. While current passage retrieval methods that focus on topical relevance have been shown to be not effective at finding answers, the result shows that more knowledge is required to identify answers in the document.
We have been formulating the answer passage extraction problem as a summarization task. We initially used term distributions extracted from a Community Question Answering (CQA) service to generate more effective summaries of retrieved web pages. An experiment was conducted to see the benefit of using the CQA data in finding answer passages. We analyze the fraction of answers covering a set of queries, the quality of the corresponding result from the answering service, and their impact on the generated summaries. I will also talk about recent work where we re-rank retrieved passages according to the summary quality and incorporating document summarizability into the ranking function.
Bio
Prof. Mark Sanderson works in Computer Science at the School Science at RMIT University in Melbourne, Australia, where he is head of the RMIT Information Retrieval (IR) group, which is regarded as the leading IR group in Australia. He is co-editor of Foundations and Trends in Information Retrieval, which is currently the highest impact rated IR journal. He is also an associate editor of IEEE TKDE and of ACM TWeb. Prof. Sanderson was co-PC chair of ACM SIGIR in 2009 and 2012, and general chair of the conference in 2004. Prof Sanderson is also a visiting professor at NII in Tokyo.
Eric Nyberg
Beyond Factoid QA: Real Questions from Real Users in Real Time
In 2011, the first IBM Watson system showed that a question-answering system could be: a) trained to be accurate and confident, and b) scaled to be fast -- enough to win at a question-answer game like Jeopardy! against human opponents in real time. Since 2011, a variety of research projects at Carnegie Mellon have focused on advancing the state of the art beyond the Jeopardy! challenge, along specific dimensions: a) automatic domain adaptation and system optimization for complex information systems (e.g., ECD, CSE); b) application of advanced QA approaches to practical domains (e.g., QUADS, BioASQ); c) generalization of QA technology to support live question-answering on the web (e.g., LiveQA). In this talk, I will present an overview of these recent results, and some ideas for future research along these dimensions.
Bio
Eric Nyberg is a Professor in the Language Technologies Institute in the School of Computer Science at Carnegie Mellon University. He is Director for the Master's Program in Computational Data Science (formerly known as the M.S. Program in Very Large Information Systems). Nyberg has made significant research contributions to the fields of automatic text translation, information retrieval, and automatic question answering. He received his Ph.D. from Carnegie Mellon University (1992), and his BA from Boston University (1983). He has pioneered the Open Advancement of Question Answering, an architecture and methodology for accelerating collaborative research in automatic question answering. In 2012, Nyberg received the Allen Newell Award for Research Excellence for his scientific contributions to the field of question answering and his work on the Watson project. He received the BU Computer Science Distinguished Alumna/Alumnus Award on September 27, 2013.
David Carmel
Web Question Answering with CQA data
In this talk I’ll cover some of our recent work on WebQA with CQA data. I’ll describe a general framework for enriching general Web search with related questions and answers extracted from CQA sites. I’ll get into the details of several components of the platform including identifying Web queries with question intent, high precision retreival from CQA data, and supporting human answers for advice seeking questions. I’ll finally overview the Trec LiveQA track which we organize the second year in a row. This track challenges participants with real-time questions extracted from the live stream of questions submitted to the Yahoo Answers site.
Bio
David is a Principal Research Scientist at Yahoo Lab at Haifa. David’s research is focused on search and content quality analysis in Web and Email, query performance prediction, vertical search, and text mining. David has published more than 100 papers in IR and Web journals and conferences, and serves on the editorial board of the IR journal and as a senior PC member or as Area Chair of many ACM conferences (SIGIR, WWW, WSDM. CIKM). He organized a number of workshops and taught several tutorials at SIGIR, and WWW. David is co-author of the book “Estimating the Query Difficulty for Information Retrieval”, published by Morgan & Claypool in 2010, and the co-author of the paper “Learning to estimate query difficulty”, which won the Best Paper Award at SIGIR 2005. David earned his PhD in Computer Science from the Technion, Israel Institute of Technology in 1997.
Scott Wen-tau Yih
Question Answering with Knowledge Bases
Building a question answering system to automatically answer natural-language questions is a long-standing research problem. While traditionally unstructured text collections are the main information source for answering questions, the development of large-scale knowledge bases provides new opportunities for open-domain factoid question answering. In this talk, I will present our recent work on semantic parsing, which maps natural language questions to structured queries that can be executed on a graph knowledge base to answer the questions. Our approach defines a query graph that resembles subgraphs of the knowledge base and can be directly mapped to a logical form. With this design, semantic parsing is reduced to query graph generation, formulated as a staged search problem. Compared to existing methods, our solution is conceptually simple and yet outperforms previous state-of-the-art results substantially.
Bio
Scott Wen-tau Yih is a Senior Researcher at Microsoft Research. His research interests include natural language processing, machine learning and information retrieval. Yih received his Ph.D. in computer science at the University of Illinois at Urbana-Champaign, where he developed the joint inference framework using integer linear programming (ILP). The approach has been widely adopted in the NLP community since then. After joining MSR in 2005, he has worked on email spam filtering, keyword extraction and search & ad relevance. His recent work focuses on continuous semantic representations using neural networks and matrix/tensor decomposition methods, with applications in lexical semantics, knowledge base embedding and question answering. Yih received the best paper award from CoNLL-2011, an outstanding paper award from ACL-2015 and has served as area chairs (HLT-NAACL-12, ACL-14, EMNLP-16), program co-chairs (CEAS-09, CoNLL-14) and action editor (Transactions of ACL) in recent years.
Evgeniy Gabrilovich
Cura Te Ipsum: answering symptom queries with question intent
About 1 percent of searches on Google are symptom-related. Those queries are rarely formulated as questions yet the question intent is often clear as users are searching for pertinent medical conditions. In this talk we will discuss the recently launched symptom search experience on Google. We use machine learning methods to identify queries with condition-seeking intent. We extract relevant health conditions by analyzing the web search results as well as by consulting the Knowledge Graph. Finally, we learn a ranker for ordering the list of relevant conditions, and evaluate the system performance with medical doctors.
Bio
Dr. Evgeniy Gabrilovich is a senior staff research scientist at Google, where he works on improving healthcare. Prior to joining Google in 2012, he was a director of research and head of the natural language processing and information retrieval group at Yahoo! Research. Evgeniy is an ACM Distinguished Scientist, and is a recipient of the 2014 IJCAI-JAIR Best Paper Prize. He is also a recipient of the 2010 Karen Sparck Jones Award for his contributions to natural language processing and information retrieval. Evgeniy is currently serving as a program chair for WWW 2017, and has served as a program chair for WSDM 2015. He earned his PhD in computer science from the Technion - Israel Institute of Technology. Recently, he graduated (with extra credit) from the Executive MD training program at Harvard Medical School.
